Unless it doesn’t accurately represent the topic, which happens, and then a researcher chooses not to read the text based on the chatbot’s summary.

Nirvana fallacy.

All these chatbots do is guess. I’m just saying a researcher might as well cut out the hallucinating middleman.